We combine geodetic, geologic, and seismic information to estimate frequencies of damaging earthquakes in three types of seismotectonic zone. Type A zones contain faults for which paleoseismic data suffice to estimate conditional probabilities. Type B zones contain faults with insufficient data for conditional probability analysis. Type C zones contain diverse or hidden faults. Each zone is assumed to have randomly distributed earthquakes plus characteristic earthquakes on specific faults. Our “cascade” model allows for multiple-segment earthquakes. Within each zone, distributed earthquakes are assumed uniform in time and space, with a truncated Gutenberg-Richter magnitude distribution. Thus, seismic hazard is defined by the characteristic earthquake rate, the rate of all distributed events, and the limiting (characteristic) magnitude. Limiting magnitudes are determined from fault lengths, while earthquake rates are determined by observed seismicity and seismic moment rate.
We present a preferred seismic hazard model with lognormal recurrence and an alternate Poissonian model. The models predict 80 to 90% probability of an m ≧ 7 earthquake within southern California before 2024. The 17 January 1994 Northridge earthquake occurred within the 13% of southern California's area having the highest moment rate density. The probability of 0.2 g or greater shaking before 2024 exceeds 60% in the Ventura and San Bernardino areas, and 50% throughout the Transverse Ranges between Santa Barbara and San Bernardino.
The predicted seismicity exceeds that observed historically. This may imply that (1) we underestimate the maximum magnitudes, (2) significant strain may be released aseismically, or (3) seismicity may have been anomalously low since 1850.